Keqiang Wang

Hotel recommendation based on user preference analysis

Recommender system offers personalized suggestions by analyzing user preference. However, the performance falls sharply when it encounters sparse data, especially meets a cold start user. Hotel is such kind of goods that suffers a lot from sparsity issue due to extremely low rating frequency. In order to handle these issues, this paper proposes a novel hotel recommendation framework. The main contribution includes: 1) We combine collaboration filtering (CF) with content-based (CBF) method to overcome sparsity issue, while ensuring high accuracy. 2) Travel intents are introduced to provide additional information for user preference analysis. 3) To provide as broad as possible recommendations, diversity techniques are employed. 4) Several experiments are conducted on the real Ctrip1 dataset, the results show that the proposed hybrid framework is competitive against classical approaches.

Time-Aware and Topic-Based Reviewer Assignment

Peer review has become the most widely-used mechanism to judge the quality of submitted papers at academic conferences or journals. However, a challenging task in peer review is to assign papers to appropriate reviewers. Both the research directions of reviewers and topics of submitted papers are often multifaceted. Besides, reviewers’ research direction may change over time and their published papers closer to current time reflect their current research direction better. Hence in this paper, we present a time-aware and topic-based reviewer assignment model. We first crawl papers published by reviewers over years from web, and then build a time-aware reviewers’ personal profile using topic model to represent the expertise of reviewers. Then the relevant degree between reviewer and submitted paper is calculated through the similarity measure. In addition, by considering statistical characteristics such as TF-IDF of the papers, the matching degree between reviewer and submitted paper is further improved. At the same time, we also consider the quality of all past reviews to measure the reviewers’ present reviews. Extensive experiments on a real-world dataset demonstrate the effectiveness of the proposed method.

Optimizing Top-k Retrieval: Submodularity Analysis and Search Strategies

The key issue in top-k retrieval — finding a set of k documents (from a large document collection) that can best answer a user’s query — is to strike the optimal balance between relevance and diversity.

Local Weighted Matrix Factorization for Implicit Feedback Datasets

Item recommendation helps people to discover their potentially interested items among large numbers of items. One most common application is to recommend items on implicit feedback datasets (e.g., listening history, watching history or visiting history). In this paper, we assume that the implicit feedback matrix has local property, where the original matrix is not globally low-rank but some sub-matrices are low-rank. In this paper, we propose Local Weighted Matrix Factorization for implicit feedback (LWMF) by employing the kernel function to intensify local property and the weight function to model user preferences. The problem of sparsity can also be relieved by sub-matrix factorization in LWMF, since the density of sub-matrices is much higher than the original matrix. We propose a heuristic method DCGASC to select sub-matrices which approximate the original matrix well. The greedy algorithm has approximation guarantee of factor \(1-\frac{1}{e}\) to get a near-optimal solution. The experimental results on two real datasets show that the recommendation precision and recall of LWMF are both improved more than 30 % comparing with the best case of WMF.

Optimizing top-k retrieval: submodularity analysis and search strategies

The key issue in top-k retrieval, finding a set of k documents (from a large document collection) that can best answer a user’s query, is to strike the optimal balance between relevance and diversity. In this paper, we study the top-k retrieval problem in the framework of facility location analysis and prove the submodularity of that objective function which provides a theoretical approximation guarantee of factor 1−

Ensemble Pruning: A Submodular Function Maximization Perspective

\frac{1}{e}

Diversifying Top-k Service Retrieval

for the (best-first) greedy search algorithm. Furthermore, we propose a two-stage hybrid search strategy which first obtains a high-quality initial set of top-k documents via greedy search, and then refines that result set iteratively via local search. Experiments on two large TREC benchmark datasets show that our two-stage hybrid search strategy approach can supersede the existing ones effectively and efficiently.

Personalized Time-aware Tag Recommendation

Item recommendation helps people to discover their potentially interested items among large numbers of items. One most common application is to recommend top-n items on implicit feedback datasets (e.g., listening history, watching history or visiting history). In this paper, we assume that the implicit feedback matrix has local property, where the original matrix is not globally low rank but some sub-matrices are low rank. In this paper, we propose Local Weighted Matrix Factorization (LWMF) for top-n recommendation by employing the kernel function to intensify local property and the weight function to model user preferences. The problem of sparsity can also be relieved by sub-matrix factorization in LWMF, since the density of sub-matrices is much higher than the original matrix. We propose a heuristic method to select sub-matrices which approximate the original matrix well. The greedy algorithm has approximation guarantee of factor